what is the difference between types atomic_int_fastN_t and atomic_int_leastN_t

Question

On cplusplus.com What I see the data types atomic_int_fastN_t and atomic_int_leastN_t where N could be 8,16,32 and 64. So what is the meaning of word least and fast in the them or how they differ? i.e. How atomic_int_fast64_t is different from at atomic_int_least64_t and atomic<int64_t> ?

Answer 1

The same principle applies as to the non-atomic typedefs: std::int_least64_t, std::int_fast32_t etc.

The "leastn" types are those whose size is at least n bits. It can be more if the implementation does not support a type of exactly n bits.

The "fastn" types are those whose size is at least n bits, and working with them does not require additional operations for the processor (i.e. working with them is fast).

For example, on a 32-bit machine, it's possible that 16-bit integers are supported, but they have to be promoted to 32-bit integers for arithmetic operations, and truncated again once computation is done. On such a machine, std::int_least16_t would be the 16-bit integer, but std::int_fast16_t wold be a 32-bit integer, since that's what is fast to work with.

The typedef std::int64_t is then guaranteed to be exactly 64 bits (and will only exist if such an integer is supported by the implementation).